CPI didn’t surprise. The bigger shock was the Fed getting dragged into politics. Data came in clean and risk bounced, the markets are still hostage to long-end yields and Fed credibility. BTC defended the floor; everything else is fighting dilution and unlocks. If yields push higher, we fade strength; if yields stay capped, we buy dips in leaders only.
Key Takeaways
- Fed independence is now a market variable. Treat it like a volatility regime, not just a headline.
- CPI was a relief, not a pivot. PPI/Retails prints are lagging; don’t overfit one release.
- Retail sales re-accelerated (+0.6% m/m). Consumerism isn’t dead; inflation narrative stays “sticky-ish,” not “solved.”
- Keep it BTC-led heading into FOMC; add beta if BTC holds post-data and rates corroborate.
- Weak long-end yields = risk-on; strong long-end yields = fade strength.
Clean CPI, Dirty Politics
CPI and PPI (November’s delayed release due to U.S shutdown) propelled this week’s volatility engine; the first domino came in line with expectations, while the second took BTC from $93k to $97k. Supercore PPI +3.5% y/y is what the Fed cares about: pausing is easy; aggressive rate cuts are harder to justify unless labor cracks.
Services were flat, but trade margins fell as retailers and wholesalers took margin pressure while goods +0.9% m/m and energy +4.6% m/m did most of the work. With headline data and core CPI not re-accelerating, the Fed can justify staying put at the Jan 27–28 meeting. However, the 3% area in producer inflation metrics (and 3.5% supercore PPI) is the reminder that inflation is plateauing above target, and not collapsing to 2% on its own.

Overall, this isn’t “inflation solved,” but rather “no upside surprise.” Inside core, BLS flags big one-offs like recreation +1.2% (largest one-month increase on record for that index) and airline fares +5.2%, while used cars -1.1% and communication -1.9%.

In data-driven weeks like this, the most common mistake is over-sizing into the number. The correct move is owning optionality into the actual release, then paying attention to the market’s reaction. If yields rip higher, crypto usually suffers, regardless of how “good” the headline may look. Next on deck is Jan 28’s FOMC meeting.

Even though the market is skeptical of rate cuts in January, the odds rise quickly after; March ~28%, April ~38%, June ~80%. The change in timing is the actionable signal. This is where the Fed Chair nomination comes into play. 2026 started with a live test of Fed independence.

Last Sunday, Powell dropped a video statement after the DOJ served the Fed grand-jury subpoenas and threatened a criminal indictment tied to his June Senate testimony about the Fed’s HQ renovation. In the video, Powell claims that the testimony angle is “pretexts,” and the real point is political pressure. He explicitly framed the indictment threat as retaliation for the Fed setting rates based on “evidence and economic conditions” rather than “the preferences of the President.”
Meanwhile, last week’s jobs report confirmed that the labor market is slowing faster than most realize. Only 50k jobs were added in December, versus 70k forecasted. But it gets worse when we look at the revisions: October revised down from -105k to -173k, and November dropped from +64k to +56k. For reference, 2025 averaged 49k in monthly gains, vs 2024’s monthly average of 168k. That’s a 71% drop in job creation YoY. And the 3-month average is now -22k per month.
Indeed, the U.S. economy looks like it’s doing just fine, but that’s not what most people are noticing. U.S. workers are now receiving their lowest share of GDP since 1947. At the same time, corporate profit margins are hitting new records. The pie is getting bigger, but workers are getting increasingly smaller slices. Productivity is accelerating, but companies are using that leverage to operate with fewer workers rather than pay existing workers more. The picture gets bleaker when AI is poured into the equation. And that wave is just getting started.
AI Digest
Crypto once promised to change money forever. Well, Bitcoin did. But Satoshi’s mystique, while helpful for story-telling, ends pretty much there. What was conceived as a form of money—unit of account, form of exchange, and store of value—has now established itself as an asset class of its own. BTC thrives more on fiat debasement than on payments or real-world adoption. And while it has spawned a whole new industry out of nowhere (RWAs, DeFi, DePIN, etc), after more than a decade we can confidently say that these developments have fallen short of meeting the expectations.

The same cannot be said of AI. AI is here to stay. It’s going nowhere. In fact, it’s only going to get better from here. Exponentially better, most likely. A new form of intelligence was handed to us with no instructions manual, and it is expected to bifurcate the economy substantially. Software becomes abundant and cheap as any form of digital production becomes commoditized; physical labor or data out of reach for LLMs gain a tailwind.
On a personal level, if you haven’t started playing around with tools that go beyond chat bots, like Claude Code, you are already late. Better late than sorry, though. Or else you will be permanently behind. It will feel uncomfortable at first, but that’s the cost of tuition; if you aren’t struggling to catch up you are surely doing something wrong. The point is leverage and exploration. Automate the tedious, accelerate workflows, and keep pushing closer to the frontier. Assuming models gain 1000x longer context windows and can sustain tasks longer than any human can, say 48h non-stop, would you agree that AGI is here?

Even with math, where skeptics claimed that LLMs would never figure out how to solve (because next-token predictors lack precise, algorithmic reasoning), AI is now making breakthrough discoveries. Since late 2025, LLMs now generate and formalize novel math proofs at scale. Recently, GPT‑5.2 Pro autonomously solved Erdos problems #728 and #729, and generated a proof for #397 that was accepted by Terence Tao. Beyond autonomy, it is AI’s speed in writing/revising proofs and formalizations, that enables more scalable, parallelized research (see AI contributions to Erdos problems).

It is unavoidable. AI is coming to fully transform multiple industries. Once you can quantify an objective and specify the process, AI systems that read everything, propose variants, and iterate at machine speed will tilt the landscape. The edge is scale: you can run “a million copies” of a competent analyst, 24/7, reading every paper and reproducing every backtest; tirelessly optimize routing and inventory management in logistics; use the continuous feedback of wearables for clinical literature; triage thousands of cases and strategy trees in the domains of law and business.

The cat is out of the box: quantification plus machine‑speed iteration changes cost curves. The only constraints are taste and verification, meaning the ability to choose worthy objectives, and proving claims with real-world implications beyond hype. Bubble or not, the world won’t be the same.
- Anthropic launched Cowork, built 100% by Claude Code as a desktop app for non-technical users.
- OpenAI launched GPT Health to track medical records and connect to health apps.
- Apple and Google signed a multi-year deal where next-gen Apple Foundation Models are based on Gemini + cloud tech; “more personalized Siri” targeted “this year.”
- Google launched Universal Commerce Protocol (UCP) + “Business Agent” for Search; checkout coming to AI Mode + Gemini for eligible listings.
- DeepSeek v4 is coming next month, aiming to outperform every OpenAI and Anthropic model in coding.
- ChatGPT will start testing Ads in its Free and Go tiers, stating that responses won’t be influenced by ads, that ads will be clearly separated and labeled, and that user conversations remain private from advertisers.
Signals Radar
BTC & Majors — If BTC holds 90k, that could be a good sign to carry selective beta on confirmation; otherwise flatten and protect ammo. Buyers defended ~$90–91k; sellers showed up into ~$93–94k.

Privacy — TheDIFC (Dubai International Financial Centre) updated its Crypto Token regime, taking effect from Jan 12. Under the new mandate, a person must not provide financial services relating to a Privacy Token or involving a Privacy Device (anything with features intended to hide, anonymize, obscure, or prevent tracing of transaction, identity, or value), approve/promote them, offer them to the public, run fund activity that invests in them, or do derivatives/instruments referencing them.

Following the delisting of many privacy assets from leading CEXs a few years ago, this didn’t catch many by surprise. A one-off case may not change the positioning of those chasing this narrative for ideological reasons. However, a major crypto hub saying “no” at the licensed exchange and custody level, means that these assets will now get even fewer compliant venues, shallower books, and higher perceived risk. That doesn’t necessarily kill the assets, but makes them more reflexive. The same headline serves as a reminder that most chains are surveilled by default, and that “real money” can’t be touched.
SUI — The chain went down on Wednesday and stalled for a couple of hours until it fully recovered. Incident aside, with ~$1B in TVL, Sui’s priority now is to grow its stablecoins market cap. The foundation sees a clear opportunity right there, just as they push the adoption of perps across ecosystem protocols. If you want to be long SUI, DEX and perps volume are the most direct proxy at the moment. Meanwhile, WAL has been acting as Sui-adjacent beta while the chain has its own mini memecoin season. For its part, Walrus plays the AI and data proxy role (in their words: “data markets for the AI era”).

HYPE vs LIT — The tough battle didn’t last long as the market is favoring scale and proven buybacks over new launch and distribution anxiety. Hyperliquid remains the market leader in volume (~$40B) and holders revenue (~$640M annualized) vs Lighter’s ~$26B in volume and ~$28M in annualized holders revenue. For HYPE, ~$23 is acting as the “hold it or lose it” level, while LIT’s weekly trading range is much broader, recently making new lows ~$2 (~$500M market cap, ~$2B FDV). The latter just rolled out LIT staking this week, which brings benefits such as premium fee tiers for market makers and high-frequency trading firms, or access to LLP (up to 10 USDC deposit into LLP for every 1 LIT staked). This will result in an increase in fees, with LIT staking allowing for fee discounts such that the lowest fee tiers remain around current levels.
ZRO — LayerZero remains trapped between “value accrual eventually” and “governance can’t turn the fee switch on” as its third referendum failed one more time for not meeting the minimum quorum despite 97% of votes in favor. Apathy is a thing, and onchain voting is proving to carry more friction than it may seem. Protocol fees remain OFF and we have to wait 6 months until the next round. Even at a point in time when buybacks were top of mind, the market leader in a winner-takes-most sector couldn’t lock such an obvious win in. In the meantime, Stargate’s revenue will keep pushing flows toward ZRO buybacks. This is the only mechanism that can counteract ongoing unlocks, of which the following takes place on Jan 20 (~25M ZRO, ~6% of released supply).

ENA — Another great example of “fee switch optionality” sitting on top of a long dilution schedule. Strong tailwinds are needed to lift this overhang.

PUMP — Cashflow floor but still a meme-adjacent casino that loses appeal as prediction markets and gambling take other forms. ~19% of the total circulating supply has been offset with revenue buybacks so far (~$250M cost).
POL — Dream setup right here: you get POL as a deflationary yield-bearing asset with airdrop optionality (Billions, Katana). This is a quantifiable catalyst that goes beyond the “2.0 vision”. First, Polygon Labs announced the Open Money Stack on Jan 9 (stablecoin payments framework), and POL caught a sharp bid on the headline. A week later they acquired Coinme and Sequence for $250M to offer regulated stablecoin payments in the U.S.

But there is more: the protocol is flipping the switch from inflationary infrastructure to a deflationary fee-driven yield asset as the chain is burning ~1M POL a day in base fees (up to 3M POL in high-activity days). Home to the largest prediction market onchain (and offchain), Polymarket, Sandeep called 2026 the year of the “POL resurrection.” That’s a net of ~3.5% in annual burn vs ~1.5% issuance from staking.
- Value driver: POL’s upside tracks Polygon Chain/Agglayer activity and fee flow.
- Transaction fees: base fee burn benefits all holders; demand spikes increase deflation.
- Priority fees: 74% to validators (stake-weighted), 26% to elected block producers.
- Staking inflation: 100M POL/year paid entirely to stakers.
- Agglayer fees: core interoperability fees designed to accrue to stakers (currently zero).
- Community growth inflation: 100M POL/year now streamlined under Polygon Foundation.
- Thesis: more users and throughput means more value to POL stakers and holders.

OP — On Jan 8, 2026, Optimism published a proposal framing OP token buybacks funded by 50% of incoming Superchain revenue, starting in February, executed monthly for the next year. This came after the token being down ~50% over the past three months and ~80% in the past year. It begs the question of whether the proposal is trying to “virtue signal” by boasting around how it has collected ~5.9k ETH in revenue. The reason for that is no other than perceived risk, as Base contributes >70% of that, and they may have an incentive to launch their own token.

JUP — For users, Jupuary’s end on Jan 30 is the key date to reap their rewards. The current Jupuary breakdown is 200M initial distribution (170M to fee-paying Jupiter users, 30M to JUP stakers), plus a 200M bonus pool to reward holding/staking through all of 2026, plus 300M reserved for Jupnet incentives (i.e., longer-duration emissions). For the protocol, however, the team is contending with the buybacks vs cost-basis question.

RAY — Launchlab’s revenue is undergoing a correction, while the protocol in aggregate annualizes ~$3M and has purchased $200M+ worth of RAY to date. That’s ~28% of the circulating supply, making it the largest % buyback of any protocol. Like JUP, it’s still another example of a protocol that is underwater on their buybacks. Nonetheless, the product suite has expanded significantly. Beyond a simple AMM, Raydium now hosts launch infrastructure, perps, RWAs, and stablecoins. Vertical integration is shaping out to be a very contested battle in the Solana ecosystem and OG players like Raydium won’t give up easily.
DRIFT — V3 went live on December; LIT’s TGE increased competition. The new liquidity layer, DLP, is expected to exit beta testing in Q1. A mobile app is also planned for release around that date. As for governance, DIP-9 laid out the case for directing protocol revenue toward development instead of indirect value accrual means such as buybacks. The proposal contemplates ~$9M in USDC collected from fees allocated to fund the next 6 months of Drift Labs expenses. This would still retain 10% of fees allocated to DLP, 10% to the insurance fund, and the remaining 80% is what would go to the protocol. The plan contemplates ~$1.5M in monthly expenses.

DBR — A big unlock relative to its market hit the market on Jan 16 (~16% of circulating). This is the kind of event that caps upside heading into specific weeks where macro takes center stage. Also worth noting that they are running 100% revenue buybacks, accumulating ~4% of the DBR supply so far with an expense of ~$9M USDC.

KMNO — Like the rest of the ecosystem, it’s undergoing a reinvention phase to stay relevant. For Kamino, that means moving beyond DeFi lending and venturing into other growth sectors like RWAs and institutional infrastructure. This involves products dedicated to fixed rates, offchain collateral, custody integrations, or private credit. See Kamino 2026, the Next Chapter.
JTO — December burned ~3.76M JTO, kickstarting a buyback sequence that ended this week targeting 11M JTO. JIP-31, however, proposes a BAM Early Adopter Subsidy that would temporarily pause JTO buybacks and redirect 100% of protocol revenue to eligible BAM validators. This is bullish for adoption/validator alignment, but bearish for short-term “buyback bid.”

PRCL — Parcel’s latest partnership with Polymarket is interesting, but not imminently tradeable yet. They will be working together to launch Real Estate prediction markets that settle against Parcl’s daily housing price indices, with Polymarket operating the markets and Parcl providing the settlement reference. The initial setup consists of straightforward city index markets (up/down over a month/quarter/year, plus threshold-style outcomes), starting with major U.S. cities and expanding in phases. Parcl also plans to explore surfacing Polymarket’s implied probabilities inside Parcl’s analytics as a “forecast” signal.

FOGO — TGE took place, debuting at $200M market cap (~$500M FDV). This is a high-performance SVM L1 built by Doug, also founder of Ambient, and a team of experienced high-frequency traders. The community airdrop bucket consists of ~6% of the genesis supply and a first distribution of 1.5% at launch. If you want exposure, the cleaner play is usually wait for the first flush + stabilization.
ZAMA — Token sale at a $55M FDV floor (8% of the supply via a sealed-bid Dutch auction running Jan 21-24 with tokens 100% unlocked at claim). This is expected to be a strong contender in the privacy race. It’s a protocol for building confidential dApps on top of any L1 or L2 using Fully Homomorphic Encryption (FHE). ZAMA is aiming to use 100% of fees for token burns, while minting new tokens as operator rewards. The protocol charges for verifying encryption proofs, decryptions, and bridging encrypted values crosschain. Fees are priced in USD but paid in $ZAMA (oracle updates the ZAMA/USD rate), to keep UX predictable and keep protocol pricing tied to usage rather than token volatility.

ONDO — Mega unlock of 1.94B ONDO (~$750M), ~57% of released supply happening on Sunday Jan 18. The allocations will go to private investors (~322M ONDO), ecosystem growth (~792M ONDO), and protocol development (825M ONDO)

EIGEN — Another big unlock awaits. Just after the Jan 1 36.82M EIGEN (~9.74% of released supply), the next cliff is around the corner: Feb 1. The constructive bit is that EigenCloud is creeping into more institutional circles as DATs stake for yield, such as SharpLink deploying ~$170M ETH to Linea via EtherFi with the explicit goal of capturing ETH staking yield + EigenCloud rewards + partner incentives. Despite narrative strength, the trade still lives or dies on unlock supply weakness vs marginal buyer strength. If you want to express bullishness, the higher-EV play is typically once that overhang clears (or hedge the date if you insist on holding).
KAITO — In the InfoFi sector, Kaito came out announcing that they will be sunsetting their Yaps platform. This comes on the back of X’s Head of Product, Nikita Bier, revoking API access from InfoFi apps that were spamming the platform. Kaito claims all this took place without prior warning. They received a legal notice only two days before X made the unilateral decision. Going forward, Kaito’s next journey will focus on more targeted marketing approaches where creators are classified into tiers based on the value they can provably deliver. They will also expand across more platforms like Youtube or TikTok.

NOISE — Noise raised a ~$7.1M seed led by Paradigm. Despite its highly anticipated launch expected to be in MegaETH, the team reverted the decision and will be going live on Base. This is another hit for the L2, as PAC also decided to release early and not wait for MEGA’s mainnet to go live. It’s a sensible move, though. Coinbase is where consumers are, and Base becomes a good fit for consumer apps. W for Base; L for MegaETH.
FUN — Sports.fun’s FUN TGE hit Jan 15. Public sale on Kraken at $0.06, the first 24h traded roughly ~$0.066 to $0.125, ending up consolidating at ~$0.108 (~$19M mkt cap on ~178M float; ~$108M FDV). To recap, the product is an onchain fantasy/predictions game where users trade athlete-shares. Initially, 40% of revenue will be allocated to token buybacks.
Conclusion
This is a week to pay for optionality, not conviction. The regime hasn’t changed: rates first, everything else second. CPI removed upside surprise, but politics just added a new risk premium to the Fed. We stay disciplined into Jan 28: if BTC holds the floor and yields cooperate, we add selectively; if yields re-accelerate, we fade rallies and cut beta fast. Keep positions liquid and time-boxed; if the trade doesn’t pay quickly, flatten and preserve ammo for the next catalyst window.


